检索结果-内蒙古大学图书馆

TSCompiler: efficient compilation framework for dynamic-shape models

science China(Information sciences) 2024年第10期67卷 67-84页

作者： Xiang LUO Chen ZHANG Chenbo GENG Yanzhi YI Jiahui HU Renwei ZHANG Zhen ZHANG Gianpietro CONSOLARO Fan YANG Tun LU Ning GU Li SHANG School of Computer Science Fudan University School of Electronic Information and Electrical Engineering Shanghai Jiao Tong University School of Computer Science and Technology Harbin Institute of Technology Huawei Technologies Co. Ltd. Huawei Paris Research Center School of Microelectronics Fudan University

Today's deep learning models face an increasing demand to handle dynamic shape tensors and computation whose shape information remains unknown at compile time and varies in a nearly infinite range at runtime. This shape dynamism brings tremendous challenges for existing compilation pipelines designed for static models which optimize tensor programs relying on exact shape values. This paper presents TSCompiler, an end-to-end compilation framework for dynamic shape models. TSCompiler first proposes a symbolic shape propagation algorithm to recover symbolic shape information at compile time to enable subsequent optimizations. TSCompiler then partitions the shape-annotated computation graph into multiple subgraphs and fine-tunes the backbone operators from the subgraph within a hardware-aligned search space to find a collection of high-performance schedules. TSCompiler can propagate the explored backbone schedule to other fusion groups within the same subgraph to generate a set of parameterized tensor programs for fused cases based on dependence analysis. At runtime, TSCompiler utilizes an occupancy-targeted cost model to select from pre-compiled tensor programs for varied tensor shapes. Extensive evaluations show that TSCompiler can achieve state-of-the-art speedups for dynamic shape models. For example, we can improve kernel efficiency by up to 3.97× on NVIDIA RTX3090, and 10.30× on NVIDIA A100 and achieve up to five orders of magnitude speedups on end-to-end latency.

关键词： machine learning tensor compilers dynamic shape operator fusion code generation auto-tuning

来源：评论

学校读者我要写书评

暂无评论

A Packet Sequence Permutation-Aware Approach to Robust Network Traffic Classification

IEEE Networking Letters

引用

IEEE Networking Letters 2024年第3期6卷 203-207页

作者： Jiang, Yanzhuo Wang, Xueman Lai, Yingxu Wang, Yipeng Beijing University of Technology College of Computer Science Beijing100124 China

Anomalies in packet length sequences caused by network topology structure and congestion greatly impact the performance of early network traffic classification. Additionally, insufficient differentiation of packet length sequences using a small number of packets also affects the performance. In this letter, we propose SePeric, a packet sequence permutation-aware approach to robust network traffic classification. By exploring the correlations within packet length sequences and adjusting them to eliminate the effects of anomalous sequence orders, as well as extracting additional features from the byte sequence of the first packet to supplement the insufficient differentiation in packet length sequences. © 2019 IEEE.

关键词： Data mining

来源：评论

学校读者我要写书评

暂无评论

Pushing one pair of labels apart each time in multi-label learning: from single positive to full labels

引用

science China(Information sciences) 2025年第6期68卷 268-285页

作者： Xiang LI Xinrui WANG Songcan CHEN MIIT Key Laboratory of Pattern Analysis and Machine Intelligence College of Computer Science and Technology/College of Artificial Intelligence Nanjing University of Aeronautics and Astronautics

In multi-label learning(MLL), it is extremely challenging to accurately annotate every appearing object due to expensive costs and limited knowledge. When facing such a challenge, a more practical and cheaper alternative should be single positive multi-label learning(SPMLL), where only one positive label needs to be provided per sample. Existing SPMLL methods usually assume unknown labels as negatives, which inevitably introduces false negatives as noisy labels. More seriously, binary cross entropy(BCE) loss is often used for training, which is notoriously not robust to noisy labels. To mitigate this issue, we customize an objective function for SPMLL by pushing only one pair of labels apart each time to suppress the domination of negative labels, which is the main culprit of fitting noisy labels in SPMLL. To further combat such noisy labels, we explore the high-rankness of the label matrix, which can also push apart different labels. By directly extending from SPMLL to MLL with full labels, a unified loss applicable to both settings is derived. As a byproduct, the proposed loss can alleviate the imbalance inherent in MLL. Experiments on real datasets demonstrate that the proposed loss not only performs more robustly to noisy labels for SPMLL but also works well for full labels. Besides, we empirically discover that high-rankness can mitigate the dramatic performance drop in SPMLL. Most surprisingly, even without any regularization or fine-tuned label correction, only adopting our loss defeats state-of-the-art SPMLL methods on CUB, a dataset that severely lacks labels.

关键词： multi-label learning single positive label noisy labels missing labels image classification

来源：评论

学校读者我要写书评

暂无评论

Event-triggered tracking control for a class of nonholonomic systems in chained form

引用

science China(Information sciences) 2023年第7期66卷 147-161页

作者： Liang XU Youfeng SU He CAI Center for Discrete Mathematics and Theoretical Computer Science Fuzhou University College of Computer and Data Science Fuzhou University School of Automation Science and Engineering South China University of Technology

In this study, the event-triggered asymptotic tracking control problem is considered for a class of nonholonomic systems in chained form for the time-varying reference input. First, to eliminate the ripple phenomenon caused by the imprecise compensation of the time-varying reference input, a novel time-varying event-triggered piecewise continuous control law and a triggering mechanism with a time-varying triggering function are developed. Second, an explicit integral input-to-state stable Lyapunov function is constructed for the time-varying closed-loop system regarding the sampling error as the external input. The origin of the closed-loop system is shown to be uniformly globally asymptotically stable for any global exponential decaying threshold signals, which in turn rules out the Zeno behavior. Moreover, infinitely fast sampling can be avoided by appropriately tuning the exponential convergence rate of the threshold signal. A numerical simulation example is provided to illustrate the proposed control approach.

关键词： event-triggered nonholonomic systems strict Lyapunov function tracking integral input-to-state stable

来源：评论

学校读者我要写书评

暂无评论

Multi-Feature Fusion Based Structural Deep Neural Network for Predicting Answer Time on Stack Overflow

引用

Journal of computer science & technology 2023年第3期38卷 582-599页

作者：郭世凯王思文李辉范玉龙刘亚清张斌 Information Science and Technology College Dalian Maritime UniversityDalian 116026China Navigation College Dalian Maritime UniversityDalian 116026China Computer Science and Technology College Shandong Technology and Business UniversityYantai 264005China

Stack Overflow provides a platform for developers to seek suitable solutions by asking questions and receiving answers on various ***,many questions are usually not answered quickly *** the questioners are eager to know the specific time interval at which a question can be answered,it becomes an important task for Stack Overflow to feedback the answer time to the *** address this issue,we propose a model for predicting the answer time of questions,named Predicting Answer Time(i.e.,PAT model),which consists of two parts:a feature acquisition and fusion model,and a deep neural network *** framework uses a variety of features mined from questions in Stack Overflow,including the question description,question title,question tags,the creation time of the question,and other temporal *** features are fused and fed into the deep neural network to predict the answer time of the *** a case study,post data from Stack Overflow are used to assess the *** use traditional regression algorithms as the baselines,such as Linear Regression,K-Nearest Neighbors Regression,Support Vector Regression,Multilayer Perceptron Regression,and Random Forest *** results show that the PAT model can predict the answer time of questions more accurately than traditional regression algorithms,and shorten the error of the predicted answer time by nearly 10 hours.

关键词： answer time structural deep neural network Stack Overflow feature acquisition feature fusion

来源：评论

学校读者我要写书评

暂无评论

Efficient Top/Bottom-k Fraction Estimation in Spatial Databases Using Bounded Main Memory

引用

Tsinghua science and technology 2022年第2期27卷 223-234页

作者： Jinbao Wang Zhuojun Duan Xixian Han Donghua Yang School of Computer Science and Technology Harbin Institute of TechnologyHarbin 150001China Computer Science Department James Madison UniversityHarrisonburgVA 22807USA

Spatial databases store objects with their locations and certain types of attached items.A variety of modern applications have been developed by leveraging the utilization of locations and items in spatial objects,such as searching points of interest,hot topics,or users’attitude in specified spatial *** many scenarios,the high and low-frequency items in a spatial region are worth noticing,considering they represent the majority’s interest or eccentric users’***,existing works have yet to identify such items in an interactive manner,despite the significance of the endeavor in decision-making *** study recognizes a novel type of analytical query,called top/bottom-k fraction query,to discover such items in spatial *** achieve fast query response,we propose a multilayered data summary that is spread out across the main memory and external memory.A memory-based estimation method for top/bottom-k fraction queries is *** maximize the use of the main memory space,we design a data summary tuning method to dynamically allocate memory space among different spatial *** proposed approach is evaluated with real-life datasets and synthetic datasets in terms of estimation *** results demonstrate the effectiveness of the proposed data summary and corresponding estimation and tuning algorithms.

关键词： exploratory analytic top-k items bottom-k items spatial database

来源：评论

学校读者我要写书评

暂无评论

A malware propagation prediction model based on representation learning and graph convolutional networks

引用

Digital Communications and Networks 2023年第5期9卷 1090-1100页

作者： Tun Li Yanbing Liu Qilie Liu Wei Xu Yunpeng Xiao Hong Liu College of Computer Science and Technology Chongqing University of Posts and TelecommunicationsChongqing400065China

The traditional malware research is mainly based on its recognition and detection as a breakthrough point,without focusing on its propagation trends or predicting the subsequently infected *** complexity of network structure,diversity of network nodes,and sparsity of data all pose difficulties in predicting *** paper proposes a malware propagation prediction model based on representation learning and Graph Convolutional Networks(GCN)to address the aforementioned ***,to solve the problem of the inaccuracy of infection intensity calculation caused by the sparsity of node interaction behavior data in the malware propagation network,a mechanism based on a tensor to mine the infection intensity among nodes is proposed to retain the network structure *** influence of the relationship between nodes on the infection intensity is also ***,given the diversity and complexity of the content and structure of infected and normal nodes in the network,considering the advantages of representation learning in data feature extraction,the corresponding representation learning method is adopted for the characteristics of infection intensity among *** can efficiently calculate the relationship between entities and relationships in low dimensional space to achieve the goal of low dimensional,dense,and real-valued representation learning for the characteristics of propagation spatial *** also design a new method,Tensor2vec,to learn the potential structural features of malware ***,considering the convolution ability of GCN for non-Euclidean data,we propose a dynamic prediction model of malware propagation based on representation learning and GCN to solve the time effectiveness problem of the malware propagation *** experimental results show that the proposed model can effectively predict the behaviors of the nodes in the network and discover the influence of different characteristics of nodes on the mal

关键词： Malware Representation learning Graph convolutional networks(GCN) Tensor decomposition Propagation prediction

来源：评论

学校读者我要写书评

暂无评论

UOUU:User-Object Distance and User-User Distance Combined Method for Collaboration Task

引用

computer Modeling in Engineering & sciences 2023年第9期136卷 3213-3238页

作者： Xiangdong Li Pengfei Wang Hanfei Xia Yuting Niu College of Computer Science and Technology Zhejiang UniversityHangzhou310000China

Augmented reality superimposes digital information onto objects in the physical world and enables multi-user *** that previous proxemic interaction research has explored many applications of user-object distance and user-user distance in an augmented reality context,respectively,and combining both types of distance can improve the efficiency of users’perception and interaction with task objects and collaborators by providing userswith insight into spatial relations of user-task object and user-user,less is concerned about howthe two types of distances can be simultaneously adopted to assist collaboration tasks *** fulfill the gap,we present UOUU,the user-object distance and user-user distance combined method for dynamically assigning tasks across *** conducted empirical studies to investigate how the method affected user collaboration tasks in terms of collaboration occurrence and overall task *** results show that the method significantly improves the speed and accuracy of the collaboration tasks as well as the frequencies of collaboration *** study also confirms the method’s effects on stimulating collaboration activities,as the UOUU method has effectively reduced the participants’perceived workload and the overall moving distances during the *** for generalising the use of the method are discussed.

关键词： Augmented reality proximity interaction computer supported cooperative work user-object distance user-user distance

来源：评论

学校读者我要写书评

暂无评论

CSNet:A Count-Supervised Network via Multiscale MLP-Mixer for Wheat Ear Counting

引用

植物表型组学（英文） 2024年第4期6卷 995-1009页

作者： Yaoxi Li Xingcai Wu Qi Wang Zhixun Pei Kejun Zhao Panfeng Chen Gefei Hao State Key Laboratory of Public Big Data College of Computer Science and TechnologyGuizhou UniversityGuiyang 550025China State Key Laboratory of Public Big Data College of Computer Science and TechnologyGuizhou UniversityGuiyang 550025China Department of Computer Science and Technology Tsinghua UniversityBeijing 100084China State Key Laboratory of Public Big Data College of Computer Science and TechnologyGuizhou UniversityGuiyang 550025China National Key Laboratory of Green Pesticide Key Laboratory of Green Pesticide and Agricultural BioengineeringMinistry of EducationGuiyang 550025China

Wheat is the most widely grown crop in the world,and its yield is closely related to global food *** number of ears is important for wheat breeding and yield ***,automated wheat ear counting techniques are essential for breeding high-yield varieties and increasing grain ***,all existing methods require position-level annotation for training,implying that a large amount of labor is required for annotation,limiting the application and development of deep learning technology in the agricultural *** address this problem,we propose a count-supervised multiscale perceptive wheat counting network(CSNet,count-supervised network),which aims to achieve accurate counting of wheat ears using quantity *** particular,in the absence of location information,CSNet adopts MLP-Mixer to construct a multiscale perception module with a global receptive field that implements the learning of small target attention maps between wheat ear *** conduct comparative experiments on a publicly available global wheat head detection dataset,showing that the proposed count-supervised strategy outperforms existing position-supervised methods in terms of mean absolute error(MAE)and root mean square error(RMSE).This superior performance indicates that the proposed approach has a positive impact on improving ear counts and reducing labeling costs,demonstrating its great potential for agricultural counting *** code is available at .

关键词： network counting csnet mlp-mixer multiscale supervised wheat

来源：评论

学校读者我要写书评

暂无评论

On learning the right attention point for feature enhancement

引用

science China(Information sciences) 2023年第1期66卷 131-143页

作者： Liqiang LIN Pengdi HUANG Chi-Wing FU Kai XU Hao ZHANG Hui HUANG College of Computer Science and Software Engineering Shenzhen University Department of Computer Science and Engineering The Chinese University of Hong Kong School of Computer Science National University of Defense Technology School of Computing Science Simon Fraser University

We present a novel attention-based mechanism to learn enhanced point features for point cloud processing tasks, e.g., classification and segmentation. Unlike prior studies, which were trained to optimize the weights of a pre-selected set of attention points, our approach learns to locate the best attention points to maximize the performance of a specific task, e.g., point cloud classification. Importantly, we advocate the use of single attention point to facilitate semantic understanding in point feature learning. Specifically,we formulate a new and simple convolution, which combines convolutional features from an input point and its corresponding learned attention point(LAP). Our attention mechanism can be easily incorporated into state-of-the-art point cloud classification and segmentation networks. Extensive experiments on common benchmarks, such as Model Net40, Shape Net Part, and S3DIS, all demonstrate that our LAP-enabled networks consistently outperform the respective original networks, as well as other competitive alternatives, which employ multiple attention points, either pre-selected or learned under our LAP framework.

关键词： point convolution feature enhancement attention point deep neural network

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：